This is the mail archive of the
binutils@sources.redhat.com
mailing list for the binutils project.
Re: PATCH: Don't count definitions in shared objects
- From: Jack Howarth <howarth at fuse dot net>
- To: "H. J. Lu" <hjl at lucon dot org>
- Cc: Franz dot Sirl-kernel at lauterbach dot com, binutils at sources dot redhat dot com
- Date: 13 Aug 2002 13:41:13 -0400
- Subject: Re: PATCH: Don't count definitions in shared objects
- References: <20020813102810.A7345@lucon.org>
HJ,
Okay. So this patch will allow glibc-2-2-branch glibc with
your proposed glibc-2-2-ppc patch to build a usable gcc 3.2, right?
I haven't started my rebuild of stock gcc 3.2 against this
new glibc build yet. Should I test your binutils patch...that is
build a new binutils before I rebuild the stock gcc 3.2 against
your proposed glibc changes? I would like to do that but I am
a little fuzzy on what I should check. Just if the resulting gcc 3.2
can pass it make check testsuite properly to make sure no
breakage has occurred?
Jack
On Tue, 2002-08-13 at 13:28, H. J. Lu wrote:
> This patch should fix the problem. I will submit a testcase patch
> also.
>
>
> H.J.
> ----
>
> 2002-08-13 H.J. Lu <hjl@gnu.org>
>
> * elflink.h (NAME(bfd_elf,size_dynamic_sections): Don't count
> definitions in shared objects when checking symbol with
> undefined version.
>
> --- bfd/elflink.h.dup Sun Aug 11 14:38:18 2002
> +++ bfd/elflink.h Tue Aug 13 10:18:31 2002
> @@ -3068,8 +3068,11 @@ NAME(bfd_elf,size_dynamic_sections) (out
> }
> free (newname);
>
> - /* Mark this version if there is a definition. */
> + /* Mark this version if there is a definition and it is
> + not defined in a shared object. */
> if (newh != NULL
> + && ((newh->elf_link_hash_flags
> + & ELF_LINK_HASH_DEF_DYNAMIC) == 0)
> && (newh->root.type == bfd_link_hash_defined
> || newh->root.type == bfd_link_hash_defweak))
> d->symver = 1;